New to LessWrong?

New Comment
43 comments, sorted by Click to highlight new comments since: Today at 7:06 PM

Good on him for trying! This'll surely be a failure of ambition, if it is one.

Q) Why not any assignments, papers or projects?

Simply put, because self-grading those things would be almost impossible. I chose final exams as the basis of evaluation because, for most classes I’ll be following, the final exam is a good basis of evaluation, and because it is relatively easy to be objective when grading.

The impression I get from most comp sci students (and the CS classes I've taken) is that the assignments are where a lot of the learning happens. If the professor of a course had most of the points coming from assignments (I do not know if this is the case for any MIT CS classes), then that suggests to me that just taking the final is an insufficient measure of proficiency.

Likewise, his serial method of "consume a class, test, repeat" seems to fly in the face of spaced repetition. If he manages to get a D in the final exam for all the classes a week after reading through the lecture notes, but then does poorly when he takes the first exam again, this suggests this may not be a useful method to learn material. (Of course, there's good reason to believe the semester model also sucks.)

I don't think he plans on not doing assignments- he just isn't grading them. Pre-exam assignments are mostly there to a) help students avoid procrastinating on learning the material and b) insure against bad luck or nerves on exam day.

Jack is mostly correct. I am planning to do assignments, however my goal is to do these later in the program where I feel the most value will come out. I've done light programming as a hobby for years, so I'm not unfamiliar with the approach, my goal is to maximize my theoretical basis of knowledge, not necessarily to become a superstar programmer (which I believe comes after years of deliberate practice, not necessarily through college anyhow). As for evaluative basis, most of the final exams I'll be writing are at least 50%, so there is at least an argument to be made that they are substantive and not peripheral to the content.

Vaniver is also correct about spaced repetition. My reason for taking this approach is to make my process more flexible early on. Once I figure out the best methods to teach myself, it will be safer to switch to doing more classes in parallel where I can get the long-term benefits of spaced repetition. As for the grading measurement, that's a completely reasonable critique. I hope to explore the tradeoffs of this approach compared to an actual MIT program and discuss that, since there will inevitably be places my methods leave weaker than a traditional program.

This'll surely be a failure of ambition, if it is one.

Young has just completed the MIT Challenge, a few days ahead of schedule. He passed the final exams and did the programming projects for all 33 classes. Read the announcement on his blog.

... Which is fucking awesome. The dude's been my inspiration for at least two years and I remember reading the announcement on his blog a year ago. In fact, it's likely that reading his blog lead me to other blogs which lead me to LessWrong. (I don't remember exactly how I found LessWrong.)

His posts on learning and time-management are very useful. The little book of productivity, an ebook he released a few years ago, is superb.

Yes! In fact I was just reading the lbod a week ago!

At MIT, some students take 8+ classes over ~15 weeks. This involves lots of busywork and an expectation of getting the highest grade (an A). [They also often do side projects.]

Scott Young aims to complete classes at the same rate. But he's skipping much of the busywork and requiring merely passing grades. I wouldn't be surprised if he pulls it off.

I'm an MIT student and currently spend 60-100 hours/class. Taking Young's approach, I could probably average 30 hours/class, which for 33 classes might be doable in about 2 months... Maybe doing 33 MIT classes in 1 month is something for a Tim Ferriss.

The pace I'm planning on sustaining (at least for the initial period) is roughly 1 class per week. I'm trying to go faster initially so I can do 2-3 weeks on later courses where I plan to do more project work.

You're absolutely right that cutting out the busywork makes my approach a lot easier than trying to do this in actual MIT classes. But that's one of the possible benefits of doing this streamlined approach to learning rather than in an institution, one of the tradeoffs I hope to discuss as the challenge progresses.

How is this done? When I was at university, scheduling conflicts would start becoming a problem at 5 classes and be almost insurmountable at 7.

Students manage conflicts by simply skipping class sessions. Last semester, I often skipped two thirds of my class sessions. As long as you read lecture notes, do the work, and show up to tests, you're fine.

Ah, okay. This is strictly forbidden elsewhere.

Whaa? How do you even manage to get signed up for conflicting classes- is MIT's registration system set up such that it allows you to do that?

In general, MIT's registration policies are "we'll provide the rope, try not to hang yourself." On the flip side, it's nearly impossible to fail out.

Living smart means exposing your ignorance. The way to be right, in the long-term, is to be shown to be wrong every day.

What a fine rationality quote.

The second sentence is. The first sentence isn't.

What do you mean? Should one try to hide their ignorance instead? I do not follow.

Yes, a lot of the time hiding ignorance is more instrumentally useful than exposing your ignorance.

The first sentence talks about "living smart". The second sentence talks about "the way to be right, in the long-term". Those are two very different things. Relatively few people have the latter as a their core life goal and it would be a stretch to say that it constitutes 'living smart' even then. (Note: I'm one of them but that changes nothing!)

I suppose instrumental depends on whom you're exposing it to. :)

I good general philosophy.

It is always more useful to expose your own ignorance to yourself, which is what the author implies, then to indulge in self-deception.

which is what the author implies

That isn't evident in the quote.

Today you are shown to be wrong. On any given day, you are shown to be wrong. By induction, you are wrong every day. If you never get to "not wrong today!" then you're not getting more rational. There's improvements (wrong every day but the questions are harder each day or something) but the line as it stands sounds superficially deep but in practice is not rationality-focused.

Not necessarily. Finding out you're incorrect about some fact of the world is a first step to uncovering a truth, indeed in the case of a dichotomy, being incorrect about a fact instructs you on the correct truth. So if you were shown to be wrong about fact A, you are almost always closer to a true belief, even if it simply the absence of a false one.

Also, being shown to be wrong every day does not mean shown to be wrong about the same thing. Each day you could be shown to be wrong about a different thing, and each error can lead to updates in your mental model for how the world works.

Although I love the pointless dissection over a single sentence, the phrase is ambiguous as most phrases are. So superficial would be the right word to describe most aphorisms, as being merely pointers to a more nuanced set of beliefs. Don't sweat the small stuff.

That's an argument against the second sentence, not the first. (So you disagree with wedrifid.)

I do too! That ... is one hell of a misreading.

edit: There are improvements that can fix the first sentence too! (exposing your ignorance to testing, to falsification, etc).

I applaud his attempt, but the 12 month goal strikes me as ridiculous; this will likely become yet another example of the planning fallacy.

Also, why is he studying physics 1 before calculus 1?

Talking before I've completed it comes off as arrogant, and that's an unfortunate tradeoff of running this challenge live. I've done as much research as I can do now, though, so the only way to try will be to actually attempt it. As for planning fallacy, I have a fairly flexible approach with a lot of backups in case some things don't work out, so that too will be discussed in my approach.

As for Physics I, I actually completed that class as my pilot test of the pace (leaving 32 to go), so if you go to the main page you can see my results. Calculus was not a prerequisite in the class, although it probably should have been, I had to make due without mastery of those concepts.

This is a great thing you're doing! I hope you succeed. (I'd count finishing in 13+ months a success, too. You're likely to find that the later courses are harder and take longer.)

Will you be making use of any nootropics as you speedrun through the courses?

I don't plan on using anything other than occasional caffeine to boost my alertness (and even then only for emergencies, as I need to sustain my pace long-term).

I wouldn't be sure about this. If you read his blog, you'll know he's completed many many projects and is probably much more aware of the planning fallacy than most people.

A lot of people do four courses over 14 weeks, and that average of 24.5 days/course makes a speed reader's ~11 days/course without all the work and stress of assignments he understands before completing unimpressive. Sounds fun though.

The pace I'm planning on sustaining is to do a class in 5 days (1 day for my work and 1 day off each week). What's impressive is all relative, I suppose, as I know plenty of people who could put my work to shame. I only hope to share in the process so people can learn from it.

From what you wrote in Holistic Learning about the use of genius and innate talent to explain away successful learning, I think we agree that anyone without some relevant disability who is in a stable environment with access to the right resources should be able to do the same, and will after we learn how to teach how to learn. By "unimpressive," I mean "what one would expect, given what the wide distribution of mental skill levels and effort made by people who complete 4-year university says about its actual difficulty and the probable level of skill and effort of the 'productivity hacking' person doing it." You are comparatively impressive, and a very special snowflake.

Are you buying the textbooks/ finding your own? Just using the video lectures (and internet for removed sections) seems unbearably slow, and you aren't in nearly as much control over the flow of information.

after we learn how to teach how to learn

Yes. After that.

Well, after that and that's successful implementation on a large scale.

What I mean is, who will teach us how to learn how to teach how to learn?

I imagine some researchers will study learners' processes for learning in terms of cognitive algorithms, mental habits, preferred thinking styles, or whatever it turns out to be that makes some people learn better and faster than others, and then experiment with ways to change the process individuals use to learn. And they'll teach us how to teach how to learn.

You are a superb straight man.

It helps that I never get the jokes.

Are you buying the textbooks/ finding your own? Just using the video lectures (and internet for removed sections) seems unbearably slow, and you aren't in nearly as much control over the flow of information.

When I was watching Khan Academy's lectures, I got good results from VLC player's time dilation; it speeds up the video and adjusts the audio's pitch to compensate, so you can adjust the pacing. I experimentally determined 1.8x to be the right speed for me, though that will depend on you, and on whose lecture you're watching, and some of the time saved should probably go into pausing the video strategically to digest things.

Yes--for my pilot course I went around 1.5-2x, strategically speeding up and slowing down. Lectures are way more efficient when you can fast-forward and rewind.